import pandas as pd import matplotlib.pyplot as plt from typing import Optional

# read database sales = pd.read_csv('data/tesco_sales.csv') competitors = pd.read_csv('data/competitor_prices.csv').dropna()

competitors["Product_id"] = competitors["Product_id"].astype('int') competitors.Date = pd.to_datetime(competitors.Date)

sales.Date = pd.to_datetime(sales.Date) sales['profit_per_item'] = sales["Selling_price"] - sales["Cost_Price"] sales['daily_profit'] = sales['profit_per_item'] * sales['Sales_qty'] gp_sales = sales.groupby("Product_id") sales.head()

competitor_products = set(competitors.Product_id.unique()) our_products = set(sales.Product_id.unique()) product_without_competitors = our_products.difference(competitor_products) product_with_competitors = our_products.intersection(competitor_products) len(product_with_competitors), len(product_without_competitors)

product_with_highest_sales = sales.groupby("Product_id").sum().sort_values("Selling_price",ascending=False)[:10].index product_with_highest_sales

for i in set(product_with_highest_sales).intersection(product_without_competitors): s1 = sales[sales.Product_id==i] plt.scatter(s1.Selling_price,s1.daily_profit) print('Product id: ', i) plt.xlabel('price') plt.ylabel('Profit') plt.show()

# for i in set(product_with_highest_sales).intersection(product_with_competitors): # s1 = sales[sales.Product_id==i] # plt.scatter(s1.Selling_price,s1.Sales_qty) # print('Product id: ', i) # plt.xlabel('price') # plt.ylabel('quantity') # plt.show()

s1['daily_profit'].cumsum().plot()

wc = 250 woc = 8 s1 = sales[sales["Product_id"]==wc].copy() s1.set_index("Date",inplace=True) s1.sort_index(inplace=True) s1['daily_profit'].cumsum().plot();plt.title("Total Profit ");plt.show() s1['Sales_qty'].plot();plt.title("# Sales");plt.show() s1['daily_profit'].plot();plt.title("Daily Profit");plt.show() s1['Selling_price'].plot();plt.title("Sell Price");plt.show() s1['Cost_Price'].plot();plt.title("CostPrice");plt.show();

s1 = sales[sales["Product_id"]==woc].copy() # s1.set_index("Sales_qty",inplace=True) s1.sort_index(inplace=True) # s1['daily_profit'].cumsum().plot();plt.title("Total Profit ");plt.show() # s1['Sales_qty'].plot();plt.title("# Sales");plt.show() # s1['daily_profit'].plot(kind='scatter');plt.title("Daily Profit");plt.show() # s1['Selling_price'].plot();plt.title("Sell Price");plt.show() # s1['Cost_Price'].plot();plt.title("CostPrice");plt.show(); s1.plot('Sales_qty', 'daily_profit',kind='scatter');plt.show() s1.plot( 'Selling_price','Sales_qty',kind='scatter')

for i, gp in competitors.groupby('Competitor_id'): break

display(gp.head()) display(sales.head())

merged = sales.merge(competitors,left_on=['Date','Product_id'], right_on=['Date','Product_id'],how='outer')

merged = merged.dropna() merged['expensive'] = merged.Selling_price > merged.Competitor_Price

merged.groupby('expensive').sum()

merged.groupby('expensive').count()

s1.sort_values('daily_profit',ascending=False).head(20)

s1.describe()

s1

s1['sp'] = s1.Selling_price.apply(round)

plt.plot(s1.groupby('sp').sum().daily_profit,'go')

plt.plot(s1.groupby('sp').mean().daily_profit,'go')

s1.groupby('sp').sum().daily_profit.sort_values()

s1.groupby('sp').mean().daily_profit.sort_values()

s1.groupby('sp').count().daily_profit.sort_values()

s1[s1.sp==83]

! pip install statsmodels pandas matplotlib seaborn

# Price optimsization # 2. Assuming we want to increase profit, provide the recommended prices for each of the products. """ # Path: read.ipynb # Price optimsization """ from statsmodels.formula.api import ols import seaborn as sns import numpy as np import pandas as pd import matplotlib.pyplot as plt from typing import Optional sns.set_style() class PriceOptimizer: def __init__( self, df: pd.DataFrame, p_id: int, competitors: Optional[pd.DataFrame] = None ) -> None: self.sales = df product_ids = df.Product_id.unique() if not p_id in product_ids: raise ValueError("Product id not in data") self.p_id = p_id self.df = df[df.Product_id == p_id].copy() self.competitors = None if competitors: competitor_ids = competitors.Product_id.unique() if p_id in competitor_ids: self.competitors = competitors[competitors.Product_id == p_id].copy() self.competitors = self.competitors.drop(columns=["Product_id"]) # most of time its a constant self.cost_price = self.df.Cost_Price.mean() self.verbose = False self.params = self.demand_curve() self.intercept = self.params.Intercept self.slope = self.params.Selling_price def demand_curve(self) -> pd.DataFrame: """ Returns the demand curve parameters """ # fit ols model model = ols(formula="Sales_qty ~ Selling_price", data=self.df).fit() # extract parameters params = model.params if self.verbose: print(model.summary()) return params def demand(self, price: float) -> float: """ Returns the demand for a given price """ return self.intercept + self.slope * price def revenue(self, p: float) -> float: """ Returns the revenue for a given quantity and selling price """ q = self.demand(p) return q * p def profit(self, p: float) -> float: """ Returns the profit for a given quantity and selling price """ q = self.demand(p) return self.revenue(p) - self.cost_price def profit_maximizer(self, plot: bool =False) -> float: """ Optimizes the price for the product """ # price range price_range = np.linspace(0, self.df.Selling_price.max()*1.2, 100) # profit for each price profit = [self.profit(p) for p in price_range] # optimal price opt_price = price_range[np.argmax(profit)] if plot: sns.lineplot(x=price_range, y=profit) # vline plt.axvline(x=opt_price, color="red") plt.title("Profit maximizer") plt.ylabel = "Profit" plt.xlabel = "Selling price" plt.show() return opt_price def plot_demand_curve(self) -> None: """ Plots the demand curve """ sns.lmplot(x="Selling_price", y="Sales_qty", data=self.df, fit_reg=True, size=4) plt.show()

sales = pd.read_csv('data/tesco_sales.csv') competitors = pd.read_csv('data/competitor_prices.csv').dropna()

po = PriceOptimizer(sales, 120) po.intercept

po.plot_demand_curve() po.profit_maximizer(plot=True)

product_ids = sales.Product_id.unique() # To see price for each product change verbose to True. verbose = False prices = {} for p_id in product_ids: po = PriceOtipmizer(sales, p_id) price = po.profit_maximizer() profit = po.profit(price) prices[p_id] = {'price': price, 'profit': profit} if verbose: print(f"Optimal Price for product #{p_id}: {price}. profit:{profit}")